Archives

October 18, 2018

Not the new oil

Ada_Lovelace_Chalon_portrait.jpg"Does data age like fish or like wine?" the economist Diane Coyle asked last week. It was one of a long list of questions she suggested researchers need to answer in a presentation at the new Ada Lovelace Institute. More important, the meeting generally asked, how can data best be used to serve the common good? The newly-created Ada Lovelace Institute is being set up to answer this sort of question.

This is a relatively new way of looking at things that has been building up over the last year or two - active rather than passive, social rather than economic, and requiring a different approach from traditional discussions of individual privacy. That might mean stewardship - management as a public good - rather than governance according to legal or quasi-legal rules; and a new paradigm for privacy, which for the last decades has been cast as an individual right rather than a social compact. As we have argued here before, it is long since time to change that last bit, a point made by Ivana Bartoletti, head of the data privacy and data protection practice for GemServ.

One of the key questions for Coyle, as an economist, is how to value data - hence the question about how it ages. In one effort, she tried to get price and volume statistics from cloud providers, and found no agreement on how they thought about their business or how they made the decision to build a new data center. Bytes are the easiest to measure - but that's not how they do it. Some thought about the number of data records, or computations per second, but these measures are insufficient without knowing the content.

"Forget 'the new oil'," she said; the characteristics are too different. Well, that's good news in a sense; if data is not the new oil then we don't have to be dinosaur bones or plankton. But given how many businesses have spent the last 20 years building their plans on the presumption that data *is* the new oil, getting them to change that view will be an uphill slog. Coyle appears willing to try: data, she said, is a public good, non-rivalrous in use, and, like many digital goods, with high fixed but low marginal costs. She went on to say, however, that personal data is not valuable, citing the small price you get if you divide Facebook's profits across its many users.

This is, of course, not really true, any more than you can decide between wine and fish: data's value depends on the beholder, the beholder's purpose, the context, and a host of other variables. The same piece of data may be valueless at times and highly valuable at others. A photograph of Brett Kavanaugh and Christine Blasey Ford on that bed in 1982, for example, would have been relatively valueless at the time, and yet be worth a fortune now, whether to suppress or to publish. The economic value might increase as long as it was kept secret - but diminish rapidly once it was made public, while the social value is zero while it's secret but huge if made public. As commodities go, data is weird. Coyle invoked Erwin Schrödinger: you don't know what you've got until you look at it. And even then, you have to keep looking as circumstances change.

That was the opening gambit, but a split rapidly surfaced in the panel, which also included Emma Prest, the executive director of DataKind. Prest and Bartoletti raised issues of consent and ethics, and data turned from a public good into a matter of human rights.

If you're a government or a large company focused on economic growth, then viewing data as a social good means wringing as much profit as you can out of it. That to date has been the direction, leading to amassing giant piles of the stuff and enabling both open and secret trades in surveillance and tracking. One often-proposed response is to apply intellectual property rights; the EU tried something like this in 1996 when it passed the Database Directive, generally unloved today, but this gives organizations rights in databases they compile. It doesn't give individuals property rights over "my" data. As tempting as IP rights might be, one problem is that a lot of data is collaboratively created. "My" medical record is a composite of information I have given doctors and their experience and knowledge-based interpretation. Shouldn't they get an ownership share?

Of course someone - probably a security someone - will be along shortly to point out that ethics, rights, and public goods are not things criminals respect. But this isn't about bad guys. Oil or not, data has always also been a source of power. In that sense, it's heartening to see that so many of these conversations - at the nascent Ada Lovelace Institute, at the St Paul's Institute PDF), at the LSE, and at Data & Society, to name just a few - are taking place. If AI is about data, robotics is at least partly about AI in a mobile substrate. Eventually, these discussions of the shape of the future public sphere will be seen for what they are: debates over the future distribution of power. Don't tell Whitehall.


Illustrations: Ada Lovelace.

Wendy M. Grossman is the 2013 winner of the Enigma Award. Her Web site has an extensive archive of her books, articles, and music, and an archive of earlier columns in this series. Stories about the border wars between cyberspace and real life are posted occasionally during the week at the net.wars Pinboard - or follow on Twitter.